Despite the central importance of circulating tumor cells (CTCs), understanding of their role in metastasis has been limited by the extreme difficulty of characterizing CTC populations over time and linking them to metastases that occur during natural tumor progression. We propose to develop a novel system that enables longitudinal and dynamic CTC studies in genetically engineered mouse models (GEMMs). The system will make possible a series of experiments that can answer fundamental questions about the relationship between CTC characteristics and metastasis: Does the time that CTCs remain in the circulatory system (half-life) change during tumor progression? Are cancer cells in primary tumors predisposed to form metastases soon after initiation or must tumors evolve to acquire metastatic potential? Do chemotherapeutic or anti-angiogenic agents promote CTC intravasation and increase the likelihood of distant metastasis? Beyond the scope of these fundamental questions, developing a system that enables liquid biopsies for dynamic and longitudinal CTC studies in GEMMs will potentiate hypothesis-driven tumor biology studies and large-scale preclinical exploration of therapeutic strategies that are not feasible in patients. Our proposed system will use mice with fluorescent reporters that brightly label all tumor cells, including CTCs a surgically installed arterio-venous shunt; and a microfluidic cell sorter controlled by a real-tie computer system. In operation, the majority of blood from a live mouse will pass through the system and return via catheter. Each time a fluorescent CTC passes through the sorter, the control system will divert a small volume of blood including the CTC to a collection port, producing a dramatically enriched CTC sample. The sample can be further concentrated and manipulated by a variety of techniques.